gh-136234: Fix _SelectorSocketTransport.writelines to be robust to connection loss #136743
Conversation
bmerry
requested review from
1st1,
asvetlov,
kumaraditya303 and
willingc
as code owners
|
All commit authors signed the Contributor License Agreement. |
|
I've previously signed the CLA on behalf of my organisation, but my work email address has changed since then (same employer though). Is there a process to reuse that form, or do I need to sign a new one? Alternatively, I can rewrite my commits to use that old email address (it still works, just deprecated). |
You need to sign again with the new email. |
Ok, will do. Might take a few days to run it past my employer. |
kumaraditya303
added
needs backport to 3.13
|
@kumaraditya303 thanks for reviewing. The CLA has had to go to our IP people (even though I've previously signed it with a different email address), so there may be a delay. I'm hoping not longer than a week or two. |
|
Okay, ping me when the CLA is done and I'll merge it, thanks! |
@kumaraditya303 I've filled in the form (had to use the Adobe eSign process since it's on behalf of an organisation). So it's now just waiting for the PSF to process it. I don't know if I get a notification when that's complete, but if I do I'll ping you again. |
|
@kumaraditya303 any idea how long it's supposed to take for a CLA signed on behalf of an organisation to be processed? I submitted the form more than 2 weeks ago and haven't seen anything about it since. |
You need to sign the CLA with the new email as well independent of organization signing. |
The form says "If you wish to sign a Contributor Agreement on behalf of an organization or to use a different Initial License, please use the manual form instead" (note the instead). Presumably it must be possible to not do the automatic form since that only caters to one of the possible initial licenses. I'll follow up with the PSF to check what's happening. |
@kumaraditya303 I've had no reply from the PSF, so for expediency I've signed the CLA as an individual so that the CLAbot will be happy. This contribution is made on behalf of my employer, for which I've completed the manual form. |
|
Thanks @bmerry for the PR, and @kumaraditya303 for merging it 🌮🎉.. I'm working now to backport this PR to: 3.13, 3.14. |
…t to connection loss (pythonGH-136743) (cherry picked from commit 7d435cf) Co-authored-by: Bruce Merry <[email protected]>
…t to connection loss (pythonGH-136743) (cherry picked from commit 7d435cf) Co-authored-by: Bruce Merry <[email protected]>
|
GH-138673 is a backport of this pull request to the 3.14 branch. |
|
GH-138674 is a backport of this pull request to the 3.13 branch. |
|
Are there any security grounds for a 3.12 backport? If an attacker can figure out how to win this race condition, they could potentially DoS a server that is not expecting an AttributeError. I guess typically that would just crash the asyncio Task that's handling the connection and the server would survive. |
…t to connection loss (pythonGH-136743) (cherry picked from commit 7d435cf) Co-authored-by: Bruce Merry <[email protected]>
…t to connection loss (python#136743)
|
GH-138702 is a backport of this pull request to the 3.14 branch. |
1 similar comment
|
GH-138702 is a backport of this pull request to the 3.14 branch. |
…t to connection loss (python#136743)
…e robust to connection loss (pythonGH-136743) (pythonGH-138702) (cherry picked from commit 5cd6cfe) Co-authored-by: Kumar Aditya <[email protected]>
2 participants
Closes #136234